DE2645018C2 - System for adaptive equalization of the amplitude response of a transmission channel - Google Patents

Description

marked by

- a first Bode network (22, F i g. 4) to shape the frequency response of a first part of the band,

- a variable gain between the first Bode network (22) and the amplifier located second Bode network (24) for shaping the frequency response of a second Part of the tape with the proviso that any Bode network has practically no shaping effect on the part of the band that is formed by the other Bode network,

- a detector circuit (36, 40, 44, 48) coupled to the output of the filter circuit for demodulating the first, second and third pilot tones,

- A reference potential source (53) and

- One to the reference potential source (53) and the detector circuit (36,40,44,48) coupled Setting circuit (52, 54, 56, 57) for setting the gain of the amplifier (26) in a more variable manner Reinforcement using the third pilot tone and / or using the first and second pilot tones for setting the gradient of the first and second Bode network in opposite directions if the first, second and third pilot tones are in their Change levels, whereby the overall amplitude response even in the presence of frequency-selective and frequency-unselective fading in the communication channel is essentially constant remain.

2. Plant according to claim 1, characterized in that

- the first Bode-Neizwerk its maximum / minimum "shock" amplitude at the frequency of the second pilot tone,

- a variable resistor control to the second Bode network for change whose frequency response is coupled is provided,

- the second Bode network has its maximum / minimum "shock" amplitude at the frequency of the second pilot tone.

- a variable resistor control that is connected to the second Bode network to change whose frequency response is coupled, is provided and

- the setting circuit (52,54,56,57) switching means to change the resistance of the control element assigned to each of the Bode networks having

3. Plant according to claim 2, characterized in that

- Each of the control elements has at least one semiconductor diode and

- each of the resistance changing circuit means a driver circuit for supplying a selectively variable direct current to the diode.

4. Plant according to claim 1, characterized in that

- the filter circuit for each pilot tone

a) a bandpass filter whose passband is centered at the corresponding pilot frequency is and

an amplifier with a flat gain curve, and
the detector circuit for each pilot tone
a) a high-frequency detector which is tuned to the frequency of the corresponding pilot tone, and

has a low-pass filter whose cutoff frequency is chosen so that the gain becomes zero in the coupling loop before the loop phase reaches 180 °.

b)

b)

The invention relates to a system for adaptive equalization of the amplitude response of a a message flow of predetermined bandwidth loaded transmission channel, which in the generic term of claim 1 specified Art.

As is well known, high frequency microwave channels are often subject to rain or fog severe signs of loss. The operation is maintained via such channels in that Sufficient power is provided on the transmit side to ensure an adequate signal-to-noise ratio on the receive side to ensure even in the presence of a strong fading, and that d> - receivers with a automatic amplification.

In addition to caused by rain i, the fog Shrinkage phenomena have been found to cause exceptional atmospheric conditions Microwave transmission over two or more distinct paths between two in sight High frequency antennas of the microwave system. Differentiate between the various signal paths typically differ in their transmission delay, so that at the receiving antenna both constructive as well as weakening interferences are made possible. If the opposite of the period of Consider the delay present in the high-frequency signal

is borrowed, then the interference can be quite selective - there can be deep zeros in parts of the band and minor changes in neighboring frequencies appear.

The change in the received power is called fading, while a Change in shrinkage as a function dt; Frequency is known as selective fading when unselective At fading, for example during heavy fog, the signal level retains its flat profile Microwave channel and simply drops in height. However, if the channel shows a selective fading, then the frequency characteristic of the channel contains one or more minima, which can be quite sharp. the The most important features of selective shrinkage are its depth and the change in response behavior in the canal. The depth of a typical selective shrinkage can be as high as 4OdB, while the Speed at which the fade occurs, up to 100 dB / second.

So far, frequency modulation has been used in practically all microwave systems is comparatively insensitive to both selective and unselective shrinkage. In more recent times it is however, considerable interest in converting existing FM systems to use Amplitude modulation, in particular single sideband modulation with suppressed carrier, emerged. In In many ways, amplitude modulation promises to be more effective than frequency modulation, but it is but in relation to the effects of a frequency suffocating Fading far more sensitive than frequency modulation.

The problem of selective shrinkage has, of course, been known to for many years lower frequencies. For example, US-PS 20 54 657 describes a circuit with which the effects selective fading can be compensated for in a shortwave radio receiver. In the described circuit, the frequency band to be controlled is divided into a number of sub-bands, one of which each is regulated separately, for example by means of one transmitted within each of the sub-bands Pilot tones. For a satisfactory regulation it is necessary that the subbands are close to each other, d. that is, they overlap each other, and this is for the known so-called multiple burst amplitude equalizers typical.

The US-PS 27 19 270 describes a somewhat more sophisticated equalizer for use in coaxial cable carrier systems. There are three adjustable equalizers connected in series with overlapping, However, differently shaped frequency characteristics controlled by three pilot tones, which are at the edges or the middle of the message band. After going through the equalizer, the Pilot tones filtered out of the news tape, amplified, detected and with a reference voltage compared. An analog computer connected to the three comparison circuits then determines the changes which must be made for the operation of the three shock equalizers to the desired To effect total equalizer adjustment.

Each of the three pilot tone feedback circuits in the equalizers of the aforesaid U.S. Patent contains a gain. This fact and the fact that the equalizer sections have overlapping frequency characteristics exposes the possibility of instability and generation of vibrations in the equalizer. This does not actually occur with the equalizer according to US Pat. No. 2,719,270, since it is designed for use in a coaxial cable system in which changes in attenuation occur slowly and are of limited amplitude around the three feedback loops are chosen so that the equalizer is always stable
Unfortunately, the circuit described in US Pat. No. 2,719,270 is not able to compensate for frequency-selective fading in a microwave system because of the extremely high speed and the comparatively large size with which such fading occurs. In order to be able to adequately equalize frequency-selective fading in the microwave range, it is necessary that the equalizer contain both a considerable amount of negative feedback and a wide bandwidth 2 »so that it can respond to the rapid changes in the signal size. The simple way described by the aforementioned US Pat.

The object of the invention is now to compensate for both the selective and the unselective To enable shrinkage in a transmission channel, even for large and fast strokes jo in particular the selective shrinkage, without instabilities and tendencies to generate vibrations to introduce.

The inventive solution to this problem is for the system of the type described in the introduction J5 characteristic of claim 1 specified.

The invention is described in detail below with reference to the drawing; it shows

F i g. 1 shows a diagram of the transmission characteristic of a fading-affected device that has been equalized according to the invention 4 »transmission channel for different fading frequencies,

F i g. 2 is a diagram showing the transmission characteristics of the transmission channel according to FIG. 1 after an additional residual equalization has been carried out 4 "> has been,

F i g. 3 is a block diagram of an exemplary microwave radio system to explain the principles of the invention,

4 shows a block diagram of an adaptive equalizer designed according to the invention,

F i g. 5 is a diagram showing the frequency response of the attenuation in the equalizer according to FIG. 4th used equalizer section and

6 shows a circuit diagram of a Bode equalization network for use in the adaptive equalizer according to FIG. 4th

The following explanations are based on a specific microwave system, e.g. B. des microwave system known as the TD-2 system, which is known to operate in the 4 GHz band with a common carrier and works with an intermediate frequency z (IF) of 70 MHz, but the invention is not limited to but also applicable to other transmission systems, for example satellite transmission systems, HF and VHF radio systems and, under certain circumstances, land line systems such as the systems working with coaxial or millimeter waveguides.

As mentioned, the invention was investigated

based on the conversion of existing microwave systems from frequency modulation to single sideband amplitude modulation with suppressed carrier. These single sideband systems (ESB) are intended for use in voice and voice band data transmission, also in data under voice band transmission (DUV) and possibly in broadband data transmission. It has been calculated that to match this wide variety of signals to a 20 MHz ESB channel operating at an intermediate frequency of 70 MHz, the maximum allowable amplitude deviation during fading is ± 2 dB with a maximum phase deviation of ± 20 degrees. In addition, so that the noise behavior of the system does not deteriorate, each adaptive equalizer used should have a noise ratio of less than 14.5 dB and a third-order intermodulation coefficient, Ma + bc, less than -63 dB. Of the various designs of equalizer that can be used for this purpose, the power series equalizer has been found to be preferred and more than adequate for equalizing speech, voiceband data and data-under-speech.

Currently, the phase lag of selective fading is not fully understood, so it is Not clear whether additional phase equalization is required to equalize more complex data signals * will. Accordingly, the present equalizer does not address this problem.

As will be discussed in detail below, the present equalizer includes an AGC circuit for frequency-independent equalization and two variable shock equalizers, which are used at each end of the channel are centered. It can be shown that the use of only two surge equalizers leads to an adaptation leads to either a linear shape or a square shape. Therefore, the inventive Equalizer successfully approximates the first three terms of a power series.

To equalize a fading microwave channel with a polynomial equalizer to ± 2 dB, it was determined that the dynamic range of the linear term must be ± 18.5 dB, and that of the quadratic terms ± 7 dB. This would require the equalizer to have two equalizer networks, which each have a dynamic range of ± 9.25 dB for the linear term and a dynamic range of Have ± 7 dB for the quadratic term, or a total dynamic range of ± 16.25 dB. An analysis of a two-way fading model used in experiments on a working microwave system actually approximates measured shrinkage data, leads however, to the conclusion that for channels that are less than 40 dB fading, large proportions of a quadratic and linear shaping is not required at the same time. This is extremely important because it means that the surge equalizer networks actually only provide a dynamic range of ± 9.25 dB to need.

In summary, the equalizer according to the invention (with the exception of the AGC circuit, circuit for automatic gain control) must be adapted to the following characteristics:

Nominal overall gain = 0 dB

Noise figure <143 dB

Ma ~ bc ■ < -63 dB

Dynamic range of each shock> ± 9.25 dB

As mentioned, high frequency microwave channels exhibit the phenomenon of both selective and unselective Shrinkage as a result of atmospheric influences. In particular, selective shrinkage is the result the reception of multiple signals that are subject to different time delays in their transmission was. A two-way fading model has been found to be observed by a large number of people Sew channel characteristics closely and therefore most useful when designing a practical equalizer is. Such a model channel has the transfer function

Hiß = 1 -> -e

/ 2-1 / / ,, Il

(D

with 2.0 log (l - also the shrinkage range in dB,

f " the frequency of the minimum shrinkage and

τ is the difference in delay between the two transmission paths.

In detail, a two-way fading with a fading depth of 2OdB, a time delay of 4 nanoseconds and a fading minimum (f _o ) at all possible frequencies provides an adequate adaptation to the amplitude characteristic of the worst fading cases, which would have to be equalized to meet the required operational quality of a typical system. That is, an equalizer that can equalize a 20 dB, 4 nanosecond fade for all positions of the fade minimum (f ₀ ) meets the system specifications. Therefore, a 20 dB, 4 nanosecond fade was used as the standard test fade in designing the equalizer described here.

As mentioned, the Bode-Shock approximation to a second-order polynomial equalizer is a simple one and effective form of equalization. The equalizer design specifically described here, i. i.e., two Bode-shock equalizers for shaping and an AGC unit for level control, was used to equalize the 2OdB-, 4 Nanosecond test fading in the 60 to 80 MHz intermediate frequency band used. After an improvement first order, in order to take advantage of the actual Bode network characteristics, were the optimal impact parameters found as follows:

Low frequency shock

High frequency shock

f _u = 61 MHz
K = 4.0979

one

/ J ₁ = 79 MHz
K = 4.6001

frequency is where the continuously variable shock has its maximum or minimum amplitude and K is a stiffness factor related to the width of the shock on the frequency axis

The equalized fading channel is shown in FIG. 1 for different points of the minimum shrinkage The AGC gain and the amplitude of the two bursts were adjusted by pushing it down to zero three points in the band determined. These points are determined by pilot tones in the vicinity of each Band limit and are located approximately in the middle of the band. Therefore, as is to be expected, shrinkage phenomena will occur with a miniriium near one of the limits of the band or near the center of the band, fine does not equalize during fading phenomena whose minima are equidistant from the pilot tones

be equalized so well. In fact, fading occurs with a minimum of 65 MHz or 75 MHz MHz range not equalized to ± 2 dB, the requirement stated above. Therefore, during the Main part of the amplitude distortion produced by the test fading is equalized by the present equalizer additional equalization will be required at the end to meet the system requirements fulfill.

As shown in Fig. 1, the resulting Error resulting from the equalization of fading effects with varying minima, only in the 65th MHz and 75 MHz ranges becomes excessive, i.e. i.e. in the approximately equidistant position from the to zero pressing pilot tones in the middle and at the limits of the 20 MHz IF band. Also lies this excessive error in the form of a smooth, bump-like curve. This specifies the use of a residual equalizer, which consists of two somewhat narrower Bode-Shock equalizers, the one at the middle are centered between the two pilot frequencies of the main equalizer. In the illustrated embodiment residual equalization pilot frequencies near 65 MHz are 75 MHz than zero uses depressing frequencies. Since shock equalizers with this level of severity are available, the Residual equalizer can be used successfully to equalize the residual error of the main equalizer. The results the combination of the main equalizer with a residual equalizer are shown in FIG. 2, and you can see that the eventual errors are within about ± 1 dB.

F i g. 3 shows, to explain the invention, a simple microwave connection with two serial connections Sections. As shown, a transmitting station 10 receives the incoming message flow, which in the considered Example of an amplitude-modulated single sideband signal with a suppressed carrier of a bandwidth of 20 MHz, centered at about 70 MHz as the intermediate frequency commonly used in the usual frequency modulation systems is used. Accordingly, the extends over the microwave link message flow to be transmitted from about 60 MHz to 80 MHz.

A resistance bridge 11 or the like is combined the incoming message flow with six low strength pilot tones driven by six pilot tone oscillators 12! -12e are generated. The combined The signal is converted in frequency to 6 GHz in the up converter 13 and then via the microwave antenna 14 to a corresponding antenna 16 at the intermediate point 17. One to the antenna 16 connected down converter 18 converts the incoming signal back to the intermediate frequency from 70 MHz. The output signal of the converter 18 is then fed to the input of an adaptive equalizer 20 supplied, which as will be discussed, to that effect is effective, a uniform signal amplitude over the 20 MHz bandwidth of the intermediate frequency band maintain even if a deeper, more frequency-selective Fading is present on the microwave link

As mentioned, the output signal /, = 61 MHz of the pilot generator 12i is close to the lower limit of the 60 to 80 MHz single sideband signal, while the output signal f ₂ = 79 MHz of the generator 12 _{2 is located} close to the upper limit of the band . In a similar way, the output signals h = 69.5 MHz and U = 70.5 MHz of the generators 12 ₃ and 12 _{4 are} symmetrical to the mid-band intermediate frequency of 70 MHz.

If, in FIG. 3, the intermediate station 17 were the terminal station -, of the system, the equalizer 20 would be that Represent component which has been referred to above as the main equalizer, since only one Section of the microwave system would be present. In reality, however, such a system has several

κι microwave intermediate stations, all of which are one Main equalizer and some also have a residual equalizer.

For the sake of simplicity, FIG. 3 a system with two serial sections in which the Intermediate frequency signal in the intermediate point 17 in an up-converter 13 'up-converted and Via a second microwave section to a receiving antenna 16 ′ at the receiving point 30 is transmitted. In the receiving station 30 sets in

; o Down converter 18 'converts the incoming signal back into the intermediate frequency range, after which it is fed to a main equalizer 20' and from there to a residual equalizer 25. Since the system shown in FIG. 3 contains a residual equalizer 25, additional pilot generators are required in the transmitting station 10 in order to add the 65 MHz and 75 MHz mid-band pilot tones required by the residual equalizer. These tones, namely / 5 and h, are generated by generators 12s and 12β, respectively.

jo F i g. 4 shows the equalizer 20 or 20 'on a larger scale Detail. The equalizer has an amplifier 21 with a flat (frequency-independent) gain curve, the output of which is connected to a Bode equalizer section 22. The outcome of the latter is with the Connected to the input of a second amplifier 23 with a flat gain curve, the output side with the input of a second Bode equalizer section 24 is connected. The output of the equalizer section 24 is at the input of an AGC amplifier 26, the output of which is connected to the input of a hybrid circuit 27 connected is. The hybrid circuit 27 has two outputs, one of which is connected to the input of one Buffer amplifier 28 and the other via an amplifier 29 to the input of a second Hybrid circuit 31 are connected. The two outputs of the hybrid circuit 31 are with the Inputs of a third and fourth hybrid circuit 32 and 33 connected. One output of the hybrid circuit 32 is connected to a bandpass filter 34, and the latter via an amplifier 35 to the input a detector circuit 36 and a low pass filter 37. Similarly, the other output is the Hybrid circuit 32 via a band pass filter 38, an amplifier 39, a detector circuit 40 with a Low pass filter 41 connected. The first output of the hybrid circuit 33 is also through a band pass filter 42, an amplifier 43, a detector circuit 44 are connected to a low-pass filter 45. Finally is too the other output of the hybrid circuit 33 via a band-pass filter 46, an amplifier 47, a detector circuit 48 connected to a low-pass filter 49.

The outputs of the low pass filters 37,41,45 and 49 are connected to the input of an analog logic circuit 52 via lines 51-1 to 51-4. A reference voltage source 53 supplies a reference potential to the analog circuit 52, which on the output side with the Input of a diode driver circuit 54 for the equalizer section 22, also with a diode driver

circuit 56 for the equalizer section 24 and finally with an AGC driver circuit 57 for the AGC amplifier 26 is connected.

It can thus be seen that the equalizer 20 is an AGC unit, two Bode surge equalizer sections and two flat gain curve amplifiers. In the present case, both the AGC unit and the equalizer sections become dynamic and continuous changed to equalize the shrinkage phenomena on the canal. As mentioned earlier, it was found that the preferred equalization strategy is to use a second order polynomial correction to approximate a push-to-zero method, which is accomplished in the illustrated embodiment is determined by finding the error at three locations in the channel and using this error signal Adjustment of the equalizer in such a way that these errors are brought to zero. The error is through Selection of a given pilot tone isolated from the channel, established and its established size with the Voltage of the reference source 53 compared. This is done for each of the pilot tones through the corresponding one Bandpass filter and the detector circuit which accordingly develops an error signal which is sent to the analog logic circuit is fed back to the corresponding equalizer section or the corresponding equalizer sections as discussed in detail below.

From Fig. 5 it can now be seen that in the illustrated embodiment four pilot tones have been inserted into the message tape. Two pilot tones (f \, / 2), which are arranged near the boundaries of the message band, are used to control the two Bode-Stoss networks. The two other pilot tones (7j, / i), which are arranged close to and symmetrically to the center of the band, are averaged out to control the gain of the AGC amplifier. If desired, a single pilot tone can also be set at the center frequency for the two mid-band pilot tones of the exemplary embodiment shown. In this case, one of the tone oscillators in FIG. 3 and one of the filter-detector branches in FIG. 4 are not applicable. As mentioned before, the additional pilot tones / 5 and fc are used to control the residual equalizer.

Fig.5 shows what as the maybe represents the most important difference of the present invention compared to the prior art, namely the The fact that the characteristics of the two Bode equalizers are chosen so that they do not differ in frequency overlap to a significant extent. This is fundamentally different from the previously known approaches different., because there the characteristics of a multiple surge equalizer have always been chosen so that they overlap each other to a considerable extent in such a way that the overall characteristic curve in the Tape has a flat course

Filters 34, 38, 42 and 46, which are advantageously crystal filters, have great influence on this dynamic behavior of the equalizer because their phase characteristics are the main cause of delay in each Are feedback loop. To maintain stability, it is necessary to assign a low-pass filter to each of the detectors in order to adjust the gain characteristic run so that the loop gain in each feedback loop reaches zero before the Loop phase reached 180 degrees The range of this The low-pass filter determines the response speed of the feedback system. To increase the response time of the equalizer, the passband of the low-pass filter must be increased and, if the stability is to be maintained, then the bandpass filters must also have a broader hand. With others ; Words, in order to maximize the response time of the equalizer, the passband of each bandpass filter must be so be as wide as possible. Other factors that make it desirable are a relatively broad one Provide passband for the bandpass filter

; i> the normal temperature effects on the passband and the nominal accuracy of ± I kHz of the pilot tones added on the transmit side. However there is conflicting considerations. In the frequency band in which the pilot tones are arranged - ideally one

1. ". No news leading part of the tape - is that normally desired signal-to-noise ratio 28 dB for a 3.3 kHz channel. When a 4OdB loss occurs, the signal-to-noise ratio deteriorates from 28 dB to only 13 dB. The energy in

: o this noise signal is allowed to pass by the bandpass filter and determined together with the pilot tones. This in turn becomes an error in the determination the pilot energy values and thus in the equalizer setting. To limit this error it is

. ·) It is therefore desirable that the passband of the Bandpass filter is made as narrow as possible. As a compromise between these two extremes was made a 3 dB bandwidth of approximately 13 kHz was chosen. The noise present in this narrow interval

jo will have a maximum error of 0.02 dB in the Cause level detection in the normal, i.e. not subject to fading, situation, and a 0.8 dB error, if there is a loss of 40 dB. Dynamic studies have shown that for

r> this bandwidth a low-pass filter with a cut-off frequency of 11 kHz to maintain stability is adequate, while at the same time maintaining an adequate speed of response. Ideally, it should Filter passband response to ± 0.05 dB at nominal ± 1 kHz range of the pilot tone must be constant in temperature. The stop band discrimination targets will be The bandpass filters 34, 38, 42 and 46 are also controlled by the presence of interfering signals should have good ability to discriminate because of the low value of the pilot tone compared to the have total channel power. For example, in a typical situation, all of the channel power is on Input of each of the crystal filters equal - 6dBm, while the level of a single pilot tone -38 dBm, which of course is some 32 dB lower. A filter with a very narrow discrimination bandwidth is necessary because of the proximity of the message band to the single-frequency pilot tones, which in individual cases can be up to 0.068%. For this reason, a 60 dB discrimination bandwidth of approx 0.13% chosen as the design goal for these filters. This reduces the error in the pilot tone level as a result interfering channel power to 0.04 dB.

Each of the in Fig. 4 shown circuit elements

Considered on its own, ω is of a completely customary type For example, those shown can be used for the crystal band-pass filters and low-pass filters like these in Reference Data for Radio Engineers, Sections 8: 41-50 and 8: 1-16,5 respectively. Edition, H. W.

Sams, Inc. New York, N.Y. the The reference voltage source and the analog logic circuit can essentially be as shown in FIGS US-PS 27 19 270 illustrated circuits correspond

while the amplifiers, diode drivers, and AGC drivers include any suitable amplifier capable of delivering the desired gain.

In theory, the dynamic range of a Bode equalizer network is twice the flat Attenuation level of the equalizer. To reach this area, the network must be terminated in the Be able to accept all values between zero and infinity. In practice, it is only possible to use a ι Realize part of this dynamic range, mainly because of parasitic reactances in the final element predominate, if this for either very small or very large resistance values is set.

Therefore, in order to realize a formation of + 1OdB for each channel end, networks with a flat attenuation curve of -15 dB required. Networks with attenuations at this level deteriorate however, the noise behavior of the system. In addition, networks that are heavily affected by attenuation suffer more of parasitic capacitances. The compromise was flat attenuation of 7.5 dB with ± 5 dB shaping chosen per equalizer network. This would result in a dynamic range of ± 1OdB four separate 7.5 dB networks may be required. Calculations performed in this equalizer structure show that the overall system gain is the same for all flat attenuation values Is zero. However, for 7.5 dB attenuation, the noise and attenuation requirements cannot be simultaneous to be met. Furthermore, any attempt to improve the noise figure degrades the value from Ma + ß-c and vice versa

However, this problem can be solved by only adding ± 5 dB shaping at each microwave intermediate station are provided. The remainder of the misalignment is then taken by the equalizer at the next intermediate station corrected. With such a configuration there are only two 7.5 dB networks at each intermediate station necessary. Also, no output buffer amplifier is used as it is necessary to measure the intermodulation noise figure to decrease the total noise figure. The final equalizer becomes a given be equipped with four ± 7.5 dB networks.

It is essential that there is a mismatch between successive intermediate stations does not degrade the overall noise performance of the system. Research has shown that, because the overall system noise during a fade increases significantly, the added noise as a result the rather moderate error alignment, 8 dB or less, is insignificant. Therefore this path seems one attractive solution to the problem.

F i g. Figure 6 shows the construction of a typical Bode equalization network of a type such as this one with success in FIG Equalizer according to Fig. 2 can be used. Each this equalization network is controlled by an electronically variable resistor element. For use in this frequency range there are various controls including field effect transistors, Suitable for PIN diodes and thermistors. Thermistors do not respond quickly enough to be selective Atrophied, and, since PIN diodes have so far been successful at very high frequencies have been used, they are preferred for the equalizer described here.

Since the nonlinear element is from the signal source and the load is separated by a frequency-selective network, the intermodulation noise with the Frequency cannot be constant. The insertion loss from nodes 1-1 'to 2-2' of the network can be easily calculated, and the intermodulation noise can be calculated using the usual noise exposure method can be measured as a function of frequency. The use of three in series Diodes instead of a single diode allow a larger bias current to have the same insertion loss to reach. This improves the linearity of the equalizer. A further improvement can be achieved are made by bridging the diodes with an additional resistor equal to the largest for the termination of the network is required.

The on in F i g. Measurements carried out in the equalization network shown in FIG. 6 show that the change in the insertion loss occurs relative to the flat attenuation level of 7.5 dB and that the intermnHulation noise M ^ + bc can be made smaller than -9OdB.

Measured noise parameters for each system component are given in the table below:

component System output level First amplifier 0 dBm (5 dB gain) -6 dBm Second amplifier 0 dBm (10 dB gain) -6 dBm Equalizer A OdBm (worst case) -6 dBm Equalizer B OdBm (worst case) -6 dBm

Noise behavior

MHz 70 MHz

80 MHz

6.93 7.19 -83.3 -84.0 8.43 8.93 -80.6 -81.5 not measured -101 not measured -101 not measured -98 not measured -98

The corresponding measured and calculated equalization system noise parameters are shown below again.

Svstemaus- Noise figure output value calculated 60 (MHz) 0 dBm 12.3 -6 dBm

80

measured 60

M λ + B-C

calculated

70

80

measured

12.5

12.9

12.2

12.6

-62.4

-62.4

Note that the A / ^ + sc term is slightly above the required -63 dB. The predominant term of this product is the M _a + bc of the amplifiers used that did not achieve the required value of -85 dB in the test equalizer that was actually built and measured . Recently, amplifiers that meet this requirement have become available, so the required -63 dB for intermodulation noise can be met.

It should be noted that the frequencies and equalizer characteristics for the embodiment described au the frequencies used in this transmission system are tailored. Obviously, too Use in other systems that work with different frequencies, appealing setting to be made for the frequencies of the pilot tones and the characteristics of the equalizer networks.

In addition 5 sheets of drawings

Claims (1)

Patent claims: 1. System for adaptive equalization of the amplitude response of a predetermined with a message flow Bandwidth of the loaded transmission channel (14,16, Fig. 3), - with a first or second located approximately at the upper and lower band limit Pilot tone (Fl or F2), - at least one third pilot tone (F3) located approximately in the middle of the band, - an amplifier (26) of variable gain and - a filter circuit (31, 32, 34, 38, 33, 42, 46) coupled to the output of the amplifier to filter out the first, second and third pilot tone from the message flow,